Subject recruitment
This work was carried out according to the principles of the Declaration of Helsinki (23) revised in 2013 for investigation with humans and following the ethical standards recognized by the Spanish biomedical research law. The Institutional Review Board in all participating medical centers approved the experimental protocol of this study and all donors signed a written informed consent.
This study included a total of 26 patients with definite MD, and 13 healthy controls, which were recruited from the Otoneurology Clinics of four academic hospitals in Spain. Over-aged individuals who fulfilled the clinical diagnosis of definite MD, according to the diagnostic criteria of the Barany Society (1), that had signed an informed consent were included in the study. Patients who suffered from another associated otological disease, such as paroxysmal positional vertigo, vestibular neuritis, vestibular schwannoma, or any other cause that could mimic MD or individuals that were under immunosuppressor, or antihistaminic treatment were excluded from the study. Clinical information from each patient was gathered during recruitment, which can be found summarized in supplementary table 5.
Blood sampling and stimulation
Peripheral blood was collected into EDTA-coated vacutainers (BD Biosciences, #367525) in the morning (8:00–13:00). Blood samples were sent to GENYO and processed within 24h of sample collection.
Whole blood was cultured under three conditions: unstimulated, stimulated with lipopolysaccharide (LPS) (InvivoGen, #tlrl-peklps), and stimulated with allergenic extract from Aspergillus niger (ASP) (DST, #42020860). So, blood was mixed 1:1 with RPMI 1640 (Thermo Fisher Scientific, #61870-044) supplemented with 1% MEM (Thermo Fisher Scientific, #11140-035) and 1% Sodium Pyruvate (Biowest, #L0642-500). If stimulation was taking place either 50 ng/mL LPS, or 10 µg/mL ASP was added, and cells were incubated for 6h at 37ºC, 7% CO2 in sterile 5 mL polystyrene round-bottom tubes (Corning, #352054). Each sample was cultured in duplicate. To one tube, a protein transport inhibitor cocktail of Brefaldin A (eBioscience, #00-4506-51) and Monensin (eBioscience #00-4505-51) was added prior to incubation, and in the other tube no protein transport inhibitor was added.
After incubation, the blood cells in the tube with protein transport inhibitor were incubated for 5 min at room temperature in 6.25 µM Cisplatin (Sigma Aldrich, #479306), for live/dead staining. Before storing the samples at -80ºC, samples were incubated for 10 min at RT with Proteomic Stabilizer (Smarttube, #PROT1-1L).
The tube with blood cells without protein transport inhibitor was centrifuged for 10 minutes at 1500 rpm, supernatant was collected and stored at -80ºC.
Mass Cytometry Immunophenotyping
For immunophenotyping, a panel of 29 metal-conjugated monoclonal antibodies was optimized and employed. All antibodies were validated, pre-titered and aliquoted into per-test amounts and stored at -80ºC. Antibodies were either purchased from Standard BioTools Inc. in pre-conjugated format or unlabelled antibodies were purchased in a carrier-protein-free format and conjugated with the indicated metal isotope using Maxpar X8 Antibody Labeling Kits (Standard BioTools Inc.), following the manufacturer’s protocol. A list of antibodies and corresponding metal tags can be seen in Table 1.
Table 1 - Mass Cytometry panel.
Lanthanide
|
Marker
|
Clone
|
Provider
|
Reference
|
RRID
|
Step
|
102Pd
|
Barcode 1
|
|
Standard BioTools Inc.
|
201060-K
|
-
|
BC
|
104Pd
|
Barcode 2
|
|
Standard BioTools Inc.
|
201060-K
|
-
|
BC
|
105Pd
|
Barcode 3
|
|
Standard BioTools Inc.
|
201060-K
|
-
|
BC
|
106Pd
|
Barcode 4
|
|
Standard BioTools Inc.
|
201060-K
|
-
|
BC
|
108Pd
|
Barcode 5
|
|
Standard BioTools Inc.
|
201060-K
|
-
|
BC
|
110Pd
|
Barcode 6
|
|
Standard BioTools Inc.
|
201060-K
|
-
|
BC
|
89Y
|
CD45
|
HI30
|
Standard BioTools Inc.
|
3089003B
|
AB_2651152
|
EX
|
141Pr
|
CD3
|
UCHT1
|
Standard BioTools Inc.
|
3141019B
|
-
|
EX
|
142Nd
|
CD19
|
HIB19
|
Standard BioTools Inc.
|
3142001B
|
AB_2651155
|
EX
|
143Nd
|
CD127
|
A019D5
|
Standard BioTools Inc.
|
3143012B
|
AB_2810240
|
EX
|
145Nd
|
CD45
|
RPA-T4
|
Standard BioTools Inc.
|
3145001B
|
AB_2661789
|
EX
|
146Nd
|
IL-1A
|
364-3B3-14
|
Biolegend
|
500108
|
AB_2890851
|
IC
|
147Sm
|
MCP1/ CCL2
|
5D3-F7
|
BD
|
550416
|
AB_2071656
|
IC
|
149Sm
|
CD66a/c/e
|
ASL-32
|
Standard BioTools Inc.
|
3149018B
|
-
|
EX
|
150Nd
|
MIP1B/ CCL4
|
D21-1351
|
Standard BioTools Inc.
|
3150004B
|
-
|
IC
|
151Eu
|
CD123
|
6H6
|
Standard BioTools Inc.
|
3151001B
|
AB_2661794
|
EX
|
152Sm
|
TNFa
|
Mab11
|
Standard BioTools Inc.
|
3152002B
|
AB_2895145
|
IC
|
153Eu
|
CD7
|
CD7-6B7
|
Standard BioTools Inc.
|
3153014B
|
AB_2811084
|
EX
|
154Sm
|
CD1c
|
L161
|
Biolegend
|
331501
|
AB_1088996
|
EX
|
155Gd
|
CD45RA
|
HI100
|
Standard BioTools Inc.
|
3155011B
|
AB_2810246
|
EX
|
156Gd
|
IL-6
|
MQ2-13A5
|
Standard BioTools Inc.
|
3156011B
|
AB_2810973
|
IC
|
158Gd
|
CD27
|
L128
|
Standard BioTools Inc.
|
3158010B
|
AB_2858231
|
EX
|
159Tb
|
CD11c
|
Bu15
|
Standard BioTools Inc.
|
3159001B
|
AB_2661800
|
EX
|
160Gd
|
IL-8
|
G265-8
|
BD
|
554720
|
AB_395529
|
IC
|
161Dy
|
IL-1B
|
CRM56
|
Thermo Fisher
|
14-7018-85
|
AB_468400
|
IC
|
163Dy
|
IL-4
|
MP4-25D2
|
Standard BioTools Inc.
|
3163011B
|
-
|
IC
|
165Ho
|
CD16
|
B73.1
|
Standard BioTools Inc.
|
3165007B
|
-
|
EX
|
165Ho
|
CD16
|
3G8
|
Standard BioTools Inc.
|
3165001B
|
AB_2802109
|
EX
|
166Er
|
IL-10
|
JES3-9D7
|
Standard BioTools Inc.
|
3166008B
|
-
|
IC
|
168Er
|
CD8
|
SK1
|
Standard BioTools Inc.
|
3168002B
|
AB_2892771
|
EX
|
169Tm
|
CD25
|
2A3
|
Standard BioTools Inc.
|
3169003B
|
AB_2661806
|
EX
|
173Yb
|
CD141
|
1A4
|
Standard BioTools Inc.
|
3173002B
|
AB_2714156
|
EX
|
174Yb
|
HLA-DR
|
L243
|
Standard BioTools Inc.
|
3174001B
|
AB_2665397
|
EX
|
175Lu
|
CD14
|
M5E2
|
Standard BioTools Inc.
|
3175015B
|
AB_2811083
|
EX
|
176Yb
|
CD56
|
N901
|
Standard BioTools Inc.
|
3176009B
|
AB_2811096
|
EX
|
191Ir/193Ir
|
DNA
|
|
Standard BioTools Inc.
|
201192B
|
-
|
DNA
|
Cis-Pt
|
|
|
Sigma
|
479306
|
-
|
live/dead
|
The panel was designed to phenotypically characterize blood cell populations, including markers for lineage, function, differentiation, and cytokines. Antibodies were divided in extra- and intracellular staining cocktails. Monoclonal antibodies that were not provided by Standard BioTools Inc. were in-house conjugated with the corresponding metal. BC- barcoding; IC – intracellular; EX – extracellular, cis-Pt - cis-Diamineplatinum(II) dichloride
Cells were stained and acquired by CyTOF as previously described (24). Briefly, whole blood samples were thawed on a roller at 4ºC for 45 min and red blood cells are lysed with thaw-lyse buffer (Smarttube, #THAWLYSE1). Cell concentration was determined, and 1.5\(\times\)106 cells were barcoded using Cell-ID 20-plex Pd Barcoding Kit (Standard BioTools Inc., #FLU201060-K). Differentially Pd-tagged samples were combined and incubated with extracellular targeted antibodies (table 4) for 30 min at 4ºC. This was followed with cell permeabilization with Perm-S buffer (Standard BioTools Inc., #FLU201066) and staining with intracellular antibodies (table 4) for 30 min at 4ºC. Cells were then washed and stained with a DNA intercalator, 0.25 µM 191Ir/193Ir (Standard BioTools Inc., #201192B) for 1h at room temperature. After this, cells were left in 2% formaldehyde (Thermo Fisher Scientific, #28906) overnight at 4ºC. Cells in the fixative solution were washed with cell staining buffer (Standard BioTools Inc., #FLU201068) and MiliQ water, cells were diluted to 5\(\times\)105 cells/mL in MiliQ water containing 1:10 diluted EQ beads (Standard BioTools Inc.) and filtered through a 100 µm mesh (Miltenyi Biotec, #130-110-917). Cells were acquired at a 250 events/second using a CyTOF 2 Helios upgraded mass cytometer (Standard BioTools Inc.). Machine tuning was performed during start-up and after 5 hours. Samples were stained and run in 40 batches. Each batch consisted of the sample conditions per patient: unstimulated, stimulated with LPS, or ASP and a batch control consisting of an unstimulated blood sample from a healthy donor, that underwent the same viability staining, fixation and storing as the study samples, and was thawed and stained in parallel with the samples in study.
Analysis of Mass Cytometry data
The .fcs files obtained from mass cytometry analysis were normalized using the processing function within the CyTOF acquisition software (version 6.7.1016) based on the run EQ four element beads.
Signal cleaning, outlier detection, file debarcoding, file aggregation, and normalization using a reference sample were carried out following the default parameters in the methods previously described (25). Only files with over 65% recovery after debarcoding were used for analysis and visualization of the generated data. This was carried out with the CyTOF workflow (26) and TreekoR (27) R packages, which perform dimensionality reduction and unsupervised clustering, manual annotation, and differential testing.
The CyTOF workflow transforms the markers intensities using an arcsinh (inverse hyperbolic sine) with cofactor 5, making the distributions more symmetric. For visualization purposes only, the data is further transformed in a 0 to 1 scale, using the 1% low percentile and the 99% high percentile as boundaries. Cell clustering, using the surface markers, was performed with FlowSOM (28) and ConsensusClusterPlus (29), which are fast methods that allow high and low frequency population identification. This was followed by a manual merging of 20 metaclusters, based on the heatmap of marker characteristics across metaclusters with dendrograms and dimensionality reduction plots (tSNE and UMAP).
Prior to the differential analysis, unknown/unassigned cells to cell populations were filtered out, using the filterSCE function from the catalyst package (30).
In the CyTOF workflow, the differential analyses use the diffcyt package (31). Various differential analyses were performed in the different categories – controls against cases and reference against treatment: (a) controls against MD, (b) individuals with low cytokines (LC) against individuals with high cytokines (HC), (c) unstimulated against stimulated with LPS, and (d) unstimulated against stimulated with Aspergillus niger.
With CyTOF workflow, DA testing was performed using a generalized linear mixed model (GLMM). With GLMM, the response variable was the cell counts per cell type and sample. The fixed effect was defined by the condition variable (disease group, treatment, or cluster). The random effect was defined by the sample ID, to model the overdispersion in proportions. A second model was used from untreated against treated comparisons, which included a random effect defined by the patient ID to account for experiment pairing.
With CyTOF workflow, for DS, the median expression of the 9 cytokines was calculated in each cell population and sample, which were used as a response variable in the linear mixed model (LMM). For absent cell populations in a sample, NAs were introduced. A filter to remove clusters with very low counts was applied. Markers with expression below 2 in at least one third of the samples were discarded from analysis.
In the analysis performed with CyTOF workflow, the p-values were corrected with a Benjamini-Hochberg adjustment using a false discovery rate (FDR) cutoff of 0.05. For DA, correction was performed for cell population and for DS it was corrected for state marker per cell population.
TreekoR was used to perform DA analysis with some modifications. FlowSOM clustering performed for the CyTOF workflow was used for TreekoR analysis. A hierarchical tree was constructed from the scaled median marker expression for each cluster, using the HOPACH method. The default value of 5 maximum children per parent node was used. For each patient, two proportions were calculated: %total and %parent. %Total refers to the proportions of cells from a cluster in each node of the tree relative to the total number of cells in the sample: (number of cells in a cluster) ÷ (number of cells in the sample). %Parent refers to the proportion of cells in each node of the tree relative to the cluster in the direct parent node of the tree (number of cells in a cluster) ÷ (number of cells in a cluster + number of cells in sibling clusters). To test if there was a significant difference between both groups, the count model EdgeR, adapted for differential abundance was used for each node in the hierarchical tree on the clusters, using %total and %parent. The cell proportions per sample were used for graphical representation of the differences of %total between groups.
Analyses were performed under R version 4.1.2.
Patient clustering
Mclust (32), a Gaussian Mixture modelling was used to obtain model-based clusters of cytokine expression. The decision regarding the number of clusters was based on the Bayesian Information criteria.
Sandwich-ELISA
Frozen supernatant samples were thawed immediately prior to analysis. IgE was measured using the commercially available IgE Human Uncoated ELISA Kit with Plates (ThermoFisher, # 88-50610-22), following the kit-specific protocols provided by the manufacturer. The absorbance was measured at 450 nm with a 570 nm correction, using the infinite m200 Nanoquant (Tecan).
Statistical analysis and visualizations
Clinical data were analysed by R, using the stats package (33). We applied a Fisher exact test for quantitative variables and Wilcoxon test for qualitative variables. Mann-Whitney U tests were performed on Sandwich-ELISA data using GraphPad Prism version 5.00 for Windows. Mann-Whitney U tests were performed on granulocyte to lymphocyte ratio data using the stats package (33). P-values below 0.05 were considered significant.
The following R packages were used for the visualizations of mass cytometry results: cytofWorkflow (26), tidyr (34), ggprism (35), ggplot2 (36), ggpubr (37) and plotrix (38).